Date post: | 10-Feb-2017 |
Category: |
Technology |
Upload: | tobias-coetzee |
View: | 141 times |
Download: | 2 times |
Background
Who am I?• Tobias Coetzee
• I’m a Technical Lead at BBD
• I present the Java Expert Level Certifications at BBD (EJB,
JPA, etc.)
• I’m currently a banker
• Part of the JoziJug organising committee
• I like to run
• This partially presentation is based on a Simon Ritter
course
• @tobiascode
• https://github.com/Jozi-JUG/streamtraining
Outcomes
What are the take away’s?• Know the differences between functional and imperative
programming
• Be able to define the elements of a stream
• Use intermediate and terminal operations
• Create streams from different data types
• Avoid NullPointerExceptions with the Optional Class
• Debug streams
Lambdas
Summary• CPU’s are getting much faster, but we are getting more cores to use.
• Inject functionality into methods, the same way we inject values into
methods.• Lambda syntax and rules, (parameters) -> {lambda body}
• Functional interfaces are interfaces that have only one abstract
method.
• Lambdas can be used where the type is a functional interface.
• Use lambdas as method parameters or assign them to variables.
• The java.util.function package already gives us a few functional
interfaces out of the box.
• Method and constructor references as short hands notation, ::.
• There are a few new methods that can use lambdas in Java 8, e.g. forEach(Consumer c).
Functional Programming
Imperative• Values associated with names can be changed, i.e. mutable state.
• The order of execution is defined as a contract.
• Repetition is external and explicit, we are responsible for the “how”.
Functional (Declarative)• Values associated with names are set once and cannot be changed,
i.e. immutable state.
• Order of execution is not defined
• Repetition is through the use of recursion or internal repetition, the
“how” is hidden away.
Thinking about a problem in terms of immutable
values and functions that translate between them.
“Multithreaded programming, requiring
synchronised access to shared, mutable
state, is the assembly language of
concurrency”
Elements of a Stream
What are streams?• Streams give us functional blocks to better process
collections of data with, monads.
• We can chain these blocks together to process
collections of data.
• Streams aren’t another data structure.
• Streams can process an infinite list of data.
• Streams use internal iteration meaning we don’t have to
code external iteration, the “how”.
• Streams support functional programming as suppose to
imperative programming.
“The purpose of streams isn’t just to convert
from one collection to another; it’s to be
able to provide a common set of operations
over data”
Elements of a Stream
Structure of a Stream• A stream consists of 3 types of things
1. A source
2. Zero or more intermediate operations
3. A terminal operation
result = albums.stream()
.filter(track -> track.getLength() > 300)
.map(Track::getName)
.collect(Collectors.toSet());
SourceIntermediate
Operation
Intermediate
Operation
Terminal
OperationResult
Elements of a Stream
How they work• The pipeline is only evaluated when the terminal
operation is called.
• The terminal operations pulls the data, the source doesn’t
push it.
• Uses the stream characteristics to help identify
optimisations.
• This allows intermediate operations to be merged:
• Avoiding multiple redundant passes on data
• Short-circuit operations
• Laze evaluation
• The stream takes care of the “how”.
Intermediate Operations
Intermediate• A stream provides a sequence of elements.
• Supports either sequential or parallel aggregate
operations.
• Most operations take a parameter that describes its
behaviour, the “what”.
• Typically using lambda expressions.
• Must be non-interfering
• Stateless
• Streams can switch between sequential and parallel, but
all processing is either done sequential or parallel, last
call wins.
Intermediate Operations
Filtering and Mapping
Interface Description
filter(Predicate) Returns a stream with only those elements that return
true for the Predicate
map(Function) Return a stream where the given Function is applied
to each element on the input stream
mapToInt,
mapToDouble,
mapToLong
Like map(), but producing streams of primitives rather
than objects.
distinct() Returns a stream with no duplicate elements
Intermediate Operations
Map() and FlatMap()• Map values from a stream either as one-to-one or one-
to-many, but still only produce one stream.
map()input stream output stream
one-to-one
flatMap()input stream output stream
one-to-many
Intermediate Operations
Sizing and Sorting
Interface Description
skip(long) Returns a stream that skips the first n elements of
the input stream
limit(long) Return a stream that only contains the first n
elements of the input stream
sorted(Comparator) Returns a stream that is sorted with the order
determined by the Comparator.
With no arguments sorts by natural order.
unordered() Returns a stream that is unordered (used internally).
Can improve efficiency of operations like distinct() and groupingBy()
Terminal Operations
Terminal• Terminates the pipeline of the operations on the
stream.
• Only at this point is any processing performed.
• This allows for optimisation of the pipeline:
• Lazy evaluation
• Merged/fused operations
• Elimination of redundant operations
• Parallel execution
• Generates an explicit result of a side effect.
Intermediate Operations
Matching Elements and Iteration
Interface Description
findFirst(Predicate) The first element that matches predicate
findAny(Predicate) Like findFirst(), but for a parallel
stream
allMatch(Predicate) All elements in the stream match predicate
anyMatch(Predicate) Any element in the stream matches predicate
noneMatch(Predicate) No elements match the predicate
forEach(Consumer) Performs an action on each element
forEachOrdered(Consumer) Like above, but ensures order is respected
when used for parallel stream
**
* Encourages non-functional programming style
Intermediate Operations
Collecting Results and Numerical Results
Interface Description
collect(Collector) Performs a mutable reduction on a stream
toArray() Returns an array containing the elements of the
stream
count() Returns how many elements are in the stream
max(Comparator) The maximum value element of the stream, returns an Optional
min(Comparator) The minimum value element of the stream, returns an Optional
average() Return the arithmetic mean of the stream, returns an Optional
sum() Returns the sum of the stream elements
*
There are a lot of built-in Collectors*
Terminal Operations
Creating a Single Result• The collect function isn’t the only option.
• reduce(BinaryOperator accumulator)
• Also called folding in FP.
• Performs a reduction on the stream using the BinaryOperator.
• The accumulator takes a partial result and the next
element and returns a new partial result as an Optional.
• Two other versions
• One that takes an initial value.• One that takes an initial value and BiFunction.
Stream Sources
JDK 8 Libraries• There are 95 methods in 23 classes that return a stream, most are
intermediate operations though.
• Leaves 71 methods in 15 classes that can be used as practical
sources.
• There are numerous places to get stream sources.
• Streams from values
• Empty streams
• Streams from functions
• Streams from arrays
• Streams from collections
• Streams from files
• Streams from other sources
Stream Sources
Collection Interface• stream()
• Provides a sequential stream of elements in the collection.• parallelStream()
• Provides a parallel stream of elements in the collection.
• Uses the fork-join framework for implementation.
• Only Collection can provide a parallel stream directly.
Arrays Class• stream()
• Array is a collection of data, so logical to be able to create a
stream.
• Provides a sequential stream.
• Overloaded methods for different types, int, double, long, Object
Stream Sources
Some other Classes• Files Class: find, list, lines, walk.
• Random numbers to produce finite or infinite streams with or without seeds: Random,
ThreadLocalRandom, SplittableRandom.
• BufferReader, JarFile, ZipFile, Pattern, Charsequence, etc.
Stream Sources
Primitive Streams• IntStream, DoubleStream, LongStream are primitive
specialisations of the Stream interface.• range(int, int), rangeClosed(int, int)
• A stream from a start to an end value (exclusive or
inclusive)• generate(IntSupplier), iterate(int,
IntUnaryOperator)
• An infinite stream created by a given Supplier.• iterate uses a seed to start the stream.
Optional
Problems with null• Certain situations in Java return a result which is a null, i.e. the
reference to an object isn’t set.• It tries to help eliminate NullPointerException’s.
• Terminal operations like min(), max(), may not return a direct result,
suppose the input stream is empty?• Introducing Optional<T>:
• It is a container for an object reference (null, or real object).
• It is a stream with either 0 or 1 elements, but never null.
• Can be used in powerful ways to provide complex conditional handling.• Doesn’t stop developers from returning null, but an Optional tells you do
maybe rather check.
• Was added for the Stream API, but you can also use it.
Optional
Creating an Optional• <T> Optional<T> empty(): Returns an empty
Optional.
• <T> Optional<T> of(T value): Returns an
Optional containing the specified value. If the
specified value is null, it throws a
NullPointerException.
• <T> Optional<T> ofNullable(T value): Same as
above, but if the specified value is null, it returns and
empty Optional.
Optional
Getting Data from an Optional• <T> get(): Returns the value, but will throw
NoSuchElementException is value is empty.
• <T> orElse(T defaultValue): Returns the default value if value
is empty.
• <T> orElseGet(Supplier<? Extends T>
defaultSupplier): Same as above, but supplier gets the value.
• <X extends Throwable> T orElseThrow(Supplier<?
Extends T> exceptionSupplier): If empty throw the
exception from the supplier.
Debugging Streams
Solution• Most IDE’s will support debugging Streams and breakpoints on
streams.
• As you can pass in the behaviour for each operation via a lambda you can always add a System.out.println() or logging as part
of the behaviour.
• If you don’t want to mess around with your current behaviour you
can use the peek operation.• The peek(Consumer <? Super T> action) methods was
added for debugging and should only be used for debugging.• Each element will pass through it and the action will be applied to
it.
Homework Exercises
How does it work?• The exercises for this month are all small exercises that you will complete
using the things you have learned during this presentation.
• There are 21 exercises to complete.
• For first month focus on using lambda expressions, method references and
some of the new methods that have been added to existing classes.
• There is a template source file that you should use to create the answers to
the exercises called Test.java under exercises package the test folder.
• To simplify things the lists and maps you need for the exercises have already
been created.
• You just need to focus on the code to solve the exercise.
• The solutions are in the Test.java file in the solutions package, but try all the
exercises first before taking a peak.